Challenges in Development and Validation of a Multiplex Assay to ...
Challenges in Development and Validation of a Multiplex Assay to Detect Biomarkers in Human Serum and Cerebrospinal Fluid 1 Jani ,
1 Plavina ,Sri
1 Laxmanan ,
1 Messinese ,
2 Stephen ,
1 Goyal
Darshana Tatiana Nicholas Laurie and Jaya Translational Sciences, Biogen Idec Inc. Cambridge, MA1 and Myriad RBM, Lake Placid, NewYork2 Abstract
Validation Results
Results and Discussion
Purpose: To develop and validate a multiplex assay in order to 1) evaluate the pharmacological activity of a biotherapeutic in human serum and cerebrospinal fluid (CSF) and 2) monitor disease activity and efficacy trends upon treatment with the biotherapeutic. Method: Using Luminex xMAP platform, a custom multiplex assay was developed to measure the concentrations of selected Biotherapeutic regulated analytes and inflammation markers (BAFF, BLC/CXCL13, MIG/CXCL9, IP-10/CXCL10, MCP-1/CCL2, MIP-3 beta/CCL19, 6Ckine/CCL21, SDF-1/CXCL12, LIGHT, ITAC/CXCL11) in human serum and CSF. For each analyte a panel of antibodies was conjugated to beads (capture) and biotin (detection). Each capture:detection antibody pair was screened by a single Luminex immunoassay using recombinant protein as a standard and human serum or CSF as assay matrix, and the best pairs selected based on the achieved sensitivity, specificity, and the ability to detect the analyte at biologically-relevant concentrations. Performance of each assay was then further optimized as a part of a 10-plex. Blinded clinical samples (serum n=24, CSF n=54) previously tested in individual ELISAs were analyzed in the developed multiplex, and based on the concordance, nine of the ten analytes were included in the final panel. The assay was then validated using standard parameters. Results: Analysis of blinded samples in a 10-plex assay demonstrated that the results obtained for all analytes correlated well with those obtained in ELISA, with the exception of LIGHT. Efforts to optimize the LIGHT assay in a multiplex format were not successful resulting in the removal of LIGHT from the panel. 9plex assay was subsequently validated in human serum and CSF demonstrating robust performance. The sensitivity and range of the individual assays were shown to be equal or superior as compared to those of corresponding ELISAs. Conclusions: A custom multiplex assay to measure concentrations of nine selected analytes was successfully developed and validated using Luminex xMAP platform. The fit-for-purpose approach to assay validation was utilized to achieve a balance between assay parameters and multiplexing capabilities, to create a practical tool for clinical studies. The assay proved to be useful in conserving assay matrices, saving time, resources and costs, and be a valuable tool for several biotherapeutic development programs.
Assay Range
Analysis of blinded samples in a 10-plex assay demonstrated that the results obtained for: •
CCL19, CCL21 and CXCL13 analytes gave good correlations to in-house ELISA results, with CCL19 exhibiting better sensitivity MIP-3 beta/CCL19, 6Ckine/CCL21 and SDF-1/CXCL12 analytes showed good correlation to the ELISA with similar or better sensitivity ITAC/CXCL11 development was also successful, having samples in the expected range LIGHT did not correlate well in a subset of samples, possibly due to differences in binding sites in the Luminex assay compared to the ELISA. A comparable pair was not found and LIGHT was removed from the multiplex No correlation between serum LIGHT concentrations derived from ELISA and 10-plex Adequate sensitivity to evaluate analytes in CSF matrix
• • •
• •
Analyte
Challenges faced during multiplex validation were that endogenous levels in serum or CSF prevented use of blank serum for specificity testing. This was mitigated by testing samples with single detection antibodies compared to multiplex and spiking a range of standard concentrations for selectivity.
Verification of CCL19, CCL21, CXCL13 and LIGHT Concentrations in the 10-plex with known blinded serum samples(n=24) y = 3.7745x R² = 0.5762
14
20
1.7
19
5.8
10
2.3
10
BAFF
9.1
20
1.6
14
20
15
8.1
13
BLC
6.3
20
1.0
9.2
33
28
13
22
IP-10
4.7
20
1.1
7.9
53
51
21
29
ITAC
2.3
20
1.0
5.3
34
35
14
13
MCP-1
5.0
20
0.90
7.8
36
21
15
23
MIG
21
20
1.6
25
36
91
15
151
MIP-3β
8.5
20
1.0
12
37
63
15
26
SDF-1
4.2
20
1.1
7.4
53
40
21
31
MQC
LQC
BAFF BLC/CXCL13 MIG/CXCL9 IP-10/CXCL10 MCP-1/CCL2 MIP-3 beta/CCL19 6Ckine/CCL21 SDF-1/CXCL12 LIGHT ITAC/CXCL11
y = 2.1006x R² = 0.0152
y = 0.2776x R² = 0.7436
Range (pg/mL) 6Ckine
Obs
MED QC
(pg/mL)
Intra CV
Inter CV
Range (pg/mL)
9.4-25
17
7%
17%
BAFF
20-45
34
10%
BLC
25-61
41
IP-10
167-283
ITAC
Dilutional Linearity
HIGH QC
(pg/mL)
Intra CV
Inter CV
Range (pg/mL)
(pg/mL)
Intra CV
Inter CV
460-884
644
4%
13%
2127-3367
2830
4%
10%
13%
680-986
816
7%
9%
14662-20773
19930
3%
5%
13%
21%
1024-1410
1287
3%
4%
3430-4888
4623
3%
4%
210
5%
10%
1577-2764
2113
5%
13%
11041-18359
15400
4%
4%
102-164
135
7%
10%
1538-2898
2149
4%
16%
4617-8928
7750
3%
8%
MCP-1
33-79
51
7%
18%
914-1210
1100
2%
6%
5873-8432
8098
6%
7%
MIG
303-668
503
9%
17%
12246-21541
14630
5%
15%
77390-131287
104660
12%
3%
MIP-3β
109-222
141
6%
11%
2371-4032
3162
3%
12%
7099-11035
9500
3%
8%
SDF-1
102-181
143
4%
19%
1590-2244
1962
2%
10%
5569-8734
8103
5%
8%
Obs
Obs
Matrix
6Ckine
of the following 10 analytes in Human Serum and
CSF
BLC
Low QC - Serum (5%) spiked with recombinant analytes, except 6Ckine
CCL19, CCL21 and CXCL13 analytes showing correlation of R2 value >0.5 to ELISA results, with CCL19 exhibiting better sensitivity No correlation between serum LIGHT concentrations derived from ELISA and 10-plex
LOW QC
Verification of BAFF and ITAC/CXCL11 in the 10-plex with known blinded serum samples(n=24) BAFF 8000
ITAC/CXCL11 1200
y = 1.091x R² = 0.830
y = 1.008x R² = 0.956
Biotinylated antibodies binds to cytokines Streptavidin-PE binds Biotinylated antibody to emit Fluorescence Fluorescence measured using Luminex 100 analyser
• • • •
Assay Development
R&D Systems ELISA (pg/mL)
R&D Systems ELISA (pg/mL)
6000 5000 4000 3000 2000
800
2000
3000
4000
5000
RBM 10-plex (pg/mL)
6000
7000
Development Determination of assay performance parameters Evaluation of assay sensitivity Evaluation of assay specificity Optimization of data reduction to express the concentration of analyte in each matrix
Inter CV
Range (pg/mL)
(pg/mL)
Intra CV
Inter CV
8.5-14
11
9%
20%
33-58
42
6%
21%
132-212
169
5%
17%
BAFF
99-121
108
3%
7%
311-417
354
4%
10%
2280-2630
2428
2%
5%
BLC
47-62
54
5%
8%
128-157
142
5%
7%
313-357
333
4%
4%
IP-10
180-230
196
2%
9%
437-486
457
2%
4%
1220-1630
1442
3%
9%
ITAC
32-49
38
7%
15%
121-146
133
2%
7%
1040-1170
1095
3%
4%
MCP-1
32-64
47
16%
20%
106-144
122
6%
9%
431-531
475
4%
7%
MIG
830-1180
969
8%
13%
2670-3560
3119
5%
10%
17300-22500
19760
2%
8%
MIP-3β
462-675
540
7%
11%
761-887
825
2%
5%
1370-1850
1577
6%
9%
SDF-1
77-98
88
7%
8%
295-336
315
3%
4%
1330-1450
1398
3%
3%
Obs
Low QC - Pooled human CSF (5%) spiked with recombinant analytes except IP-10 and MCP-1
CSF
Serum
CSF
Serum
0
200
400
600
800
1000
1200
RBM 10-plex (pg/mL)
CXCL13 was not detected in the ELISA but was detected in the multiplex assay
Nine of ten analytes were scaled up for method validation. Validation was performed for plasma, serum and CSF matrices.
Serum
CSF
Selectivity
Serum
CSF
Serum (n=3)
Plasma (n=3)
CSF (n=3)
Dilution 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16
Exp (pg/mL) 216 104 52 58 29 15 1034 517 258 54 27 13 95 48 58 1079 539 270 1079 539 270 862 431 216 4.6 8325 4163 2081 159 80 40 18350 9175 4588 418 209 104 788 394 197 259 129 65 3848 1924 962 453 226 113
Obs (pg/mL) 203 101 51 58 34 17 1030 549 249 48 22 12 112 63 65 1265 690 350 1265 690 350 862 467 262 3.9 9075 4470 2315 191 89 43 19850 10170 4615 360 200 111 808 357 161 255 133 53 3845 1945 977 492 287 158
Recov 107% 103% 103% 100% 116% 114% 100% 106% 96% 90% 82% 91% 117% 132% N/A 112% N/A N/A 117% 128% 130% 117% 128% 130% 100% 108% 122% 86% N/A N/A 109% 107% 111% 120% 111% 107% 108% 111% 101% 86% 96% 106% 103% 91% 82% 99% 102% 81% 100% 101% 102% 109% 127% 140%
Results: - Clinical samples will be run at an MRD of 1:5 for serum/plasma and 1:2 for CSF, unless a High reading is obtained for an analyte and it requires further dilution (up to the maximal dilution). - Serum and plasma samples can be maximally diluted up to 1:20-1:40 depending upon the analyte, whereas CSF samples can be diluted up to 1:4-1:16.
Spike Level Analytes
1
2
3
1
2
3
1
2
3
6Ckine
85
92
68
118
69
36
96
87
74
BAFF
115
119
102
103
108
97
102
96
104
BLC
122
124
118
94
113
108
81
88
96
IP-10
82
85
83
52
62
59
101
105
104
ITAC
94
92
94
58
66
65
103
104
101
MCP-1
88
89
87
82
97
88
105
104
93
MIG
86
87
77
86
84
86
97
93
100
MIP-3β
132
128
115
92
102
93
81
88
85
SDF-1
63
67
70
41
50
53
116
112
106
Results: CSF recovered all the spiked in concentrations for all the analytes. Serum passed for all recoveries except for SDF-1. Plasma had low recovery when spiking for 6Ckine, IP-10, ITAC, and SDF-1. h, 2- Mid, 3Greater sensitivity in 10-plex compared to ELISA
Serum
Controls were measured in duplicate
Reagent Selection ELISA Kits vs Singleplex and multiplex Luminex reagents Optimization of assay matrix Selection of assay format Selection of assay controls
(pg/mL)
Intra CV
Obs
Medium QC - 1:3.4 dilution of high QC
• Excellent correlation of 10-plex to the ELISA with similar sensitivity Correlation between ELISA and 10-plex derived CCL19 & CXCL13 concentrations in human CSF
Assay range Precision Selectivity Sensitivity LDD & LLOQ Dilution linearity Martix Interference Sample Stability
Range (pg/mL)
High QC - Pooled human CSF (25%) spiked with recombinant analytes except IP-10 and MCP-1
Validation • • • • • • • •
Inter CV
400
0 1000
(pg/mL)
Intra CV
CSF
200
0
6Ckine
Obs
HIGH QC
600
1000 0
Range (pg/mL)
MED QC
CSF
ITAC
•
Precision of CSF-based Controls
MCP-1
•
IP-10
Serum
Add 5ul capture antibody beads + 5ul blocking buffer + 10uL standard/control/sample in 96 well plate Incubate 1 hr at ambient temp Add 10uL biotinylated detection antibody Incubate 1 hr at ambient temp Add 10uL Streptavidin-PE Incubate 1 hr at ambient temp Transfer contents to Pre wetted filter membrane 96 well plate 2x wash with 100uL wash buffer FL intensity read in Luminex 100 analyser Data analysis by Plateviewer software using Best fit option MFI interpolated using standard curve
• • • • • •
CSF
Controls were measured in duplicate
Antibodies for each analyte were purchased and conjugated to beads or biotinylated (detection) using standard protocols. Each bead and detection was screened by Luminex assay with a recombinant protein and serum samples. Pairs were selected based on sensitivity, linearity of dilution and the ability to detect the samples at expected levels, followed by multiplexing in a 10plex assay. Assay verification for MIP-3 beta/CCL19, 6Ckine/CCL21, SDF-1/CXCL12, LIGHT BLC/CXCL13 was performed by running blinded samples in the 9-plex with known values. BAFF and ITAC/CXCL11 were correlated to ELISA. MIG/CXCL9, IP-10/CXCL10 and MCP-1/CCL2 were previously used in multiplex assay and did not undergo further verification. Nine of ten analytes where then scaled up for final validation. The assay was validated for Least Detectable Dose(LDD), lower limit of quantitation (LLOQ), spike-recovery, linearity and precision. Validation was performed for serum and CSF.
Cytokines bind to antibodies
CSF
Serum
Medium QC - Serum (98%) spiked with recombinant analytes except 6Ckine, BAFF, and SDF-1
•
Serum
Serum
High QC - Serum (97%) spiked with all recombinant analytes
Micrspheres coated with antibody
• • •
6Ckine
LOW QC
1000
•
LLOQ (pg/mL)
y = 0.0276x R² = 0.6422
7000
• • •
LDD (pg/mL)
Precision of Serum-based Controls
Assay Format
•
LLOQ (pg/mL)
MIG
•
LDD* (pg/mL)
MIP-3β
•
HQC
SDF-1
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
SD
CSF
BAFF
10-plex Assay Development consisting Cerebrospinal fluid:
n
Serum/Plasma
Mean + 3SD (MFI)
*Least Detectable Dose (LDD) The LDD characterizes the sensitivity of the kit and is determined by adding three standard deviations to the average signal of 20 replicate determinations of standard curve blanks and calculating the corresponding concentration.
Methods •
Mean MFI
Representative Standard Curves
Conclusion Analytical validation supported the use of multiplex assay as a valuable tool for clinical studies to evaluate the pharmacological activity of biotherapeutic in human serum and CSF or monitor disease activity and efficacy trends upon treatment with biotherapeutic. Results from the multiplex assay development matched reference ELISA assay results with equal or greater sensitivity for 9 out of 10 assays The 9-plex assay was shown to be accurate and specific in both serum and CSF matrices The assay is rapid and reproducible and allows for the quantitative measure of 9 analytes in a single 50 µl sample
1 Plavina ,Sri
1 Laxmanan ,
1 Messinese ,
2 Stephen ,
1 Goyal
Darshana Tatiana Nicholas Laurie and Jaya Translational Sciences, Biogen Idec Inc. Cambridge, MA1 and Myriad RBM, Lake Placid, NewYork2 Abstract
Validation Results
Results and Discussion
Purpose: To develop and validate a multiplex assay in order to 1) evaluate the pharmacological activity of a biotherapeutic in human serum and cerebrospinal fluid (CSF) and 2) monitor disease activity and efficacy trends upon treatment with the biotherapeutic. Method: Using Luminex xMAP platform, a custom multiplex assay was developed to measure the concentrations of selected Biotherapeutic regulated analytes and inflammation markers (BAFF, BLC/CXCL13, MIG/CXCL9, IP-10/CXCL10, MCP-1/CCL2, MIP-3 beta/CCL19, 6Ckine/CCL21, SDF-1/CXCL12, LIGHT, ITAC/CXCL11) in human serum and CSF. For each analyte a panel of antibodies was conjugated to beads (capture) and biotin (detection). Each capture:detection antibody pair was screened by a single Luminex immunoassay using recombinant protein as a standard and human serum or CSF as assay matrix, and the best pairs selected based on the achieved sensitivity, specificity, and the ability to detect the analyte at biologically-relevant concentrations. Performance of each assay was then further optimized as a part of a 10-plex. Blinded clinical samples (serum n=24, CSF n=54) previously tested in individual ELISAs were analyzed in the developed multiplex, and based on the concordance, nine of the ten analytes were included in the final panel. The assay was then validated using standard parameters. Results: Analysis of blinded samples in a 10-plex assay demonstrated that the results obtained for all analytes correlated well with those obtained in ELISA, with the exception of LIGHT. Efforts to optimize the LIGHT assay in a multiplex format were not successful resulting in the removal of LIGHT from the panel. 9plex assay was subsequently validated in human serum and CSF demonstrating robust performance. The sensitivity and range of the individual assays were shown to be equal or superior as compared to those of corresponding ELISAs. Conclusions: A custom multiplex assay to measure concentrations of nine selected analytes was successfully developed and validated using Luminex xMAP platform. The fit-for-purpose approach to assay validation was utilized to achieve a balance between assay parameters and multiplexing capabilities, to create a practical tool for clinical studies. The assay proved to be useful in conserving assay matrices, saving time, resources and costs, and be a valuable tool for several biotherapeutic development programs.
Assay Range
Analysis of blinded samples in a 10-plex assay demonstrated that the results obtained for: •
CCL19, CCL21 and CXCL13 analytes gave good correlations to in-house ELISA results, with CCL19 exhibiting better sensitivity MIP-3 beta/CCL19, 6Ckine/CCL21 and SDF-1/CXCL12 analytes showed good correlation to the ELISA with similar or better sensitivity ITAC/CXCL11 development was also successful, having samples in the expected range LIGHT did not correlate well in a subset of samples, possibly due to differences in binding sites in the Luminex assay compared to the ELISA. A comparable pair was not found and LIGHT was removed from the multiplex No correlation between serum LIGHT concentrations derived from ELISA and 10-plex Adequate sensitivity to evaluate analytes in CSF matrix
• • •
• •
Analyte
Challenges faced during multiplex validation were that endogenous levels in serum or CSF prevented use of blank serum for specificity testing. This was mitigated by testing samples with single detection antibodies compared to multiplex and spiking a range of standard concentrations for selectivity.
Verification of CCL19, CCL21, CXCL13 and LIGHT Concentrations in the 10-plex with known blinded serum samples(n=24) y = 3.7745x R² = 0.5762
14
20
1.7
19
5.8
10
2.3
10
BAFF
9.1
20
1.6
14
20
15
8.1
13
BLC
6.3
20
1.0
9.2
33
28
13
22
IP-10
4.7
20
1.1
7.9
53
51
21
29
ITAC
2.3
20
1.0
5.3
34
35
14
13
MCP-1
5.0
20
0.90
7.8
36
21
15
23
MIG
21
20
1.6
25
36
91
15
151
MIP-3β
8.5
20
1.0
12
37
63
15
26
SDF-1
4.2
20
1.1
7.4
53
40
21
31
MQC
LQC
BAFF BLC/CXCL13 MIG/CXCL9 IP-10/CXCL10 MCP-1/CCL2 MIP-3 beta/CCL19 6Ckine/CCL21 SDF-1/CXCL12 LIGHT ITAC/CXCL11
y = 2.1006x R² = 0.0152
y = 0.2776x R² = 0.7436
Range (pg/mL) 6Ckine
Obs
MED QC
(pg/mL)
Intra CV
Inter CV
Range (pg/mL)
9.4-25
17
7%
17%
BAFF
20-45
34
10%
BLC
25-61
41
IP-10
167-283
ITAC
Dilutional Linearity
HIGH QC
(pg/mL)
Intra CV
Inter CV
Range (pg/mL)
(pg/mL)
Intra CV
Inter CV
460-884
644
4%
13%
2127-3367
2830
4%
10%
13%
680-986
816
7%
9%
14662-20773
19930
3%
5%
13%
21%
1024-1410
1287
3%
4%
3430-4888
4623
3%
4%
210
5%
10%
1577-2764
2113
5%
13%
11041-18359
15400
4%
4%
102-164
135
7%
10%
1538-2898
2149
4%
16%
4617-8928
7750
3%
8%
MCP-1
33-79
51
7%
18%
914-1210
1100
2%
6%
5873-8432
8098
6%
7%
MIG
303-668
503
9%
17%
12246-21541
14630
5%
15%
77390-131287
104660
12%
3%
MIP-3β
109-222
141
6%
11%
2371-4032
3162
3%
12%
7099-11035
9500
3%
8%
SDF-1
102-181
143
4%
19%
1590-2244
1962
2%
10%
5569-8734
8103
5%
8%
Obs
Obs
Matrix
6Ckine
of the following 10 analytes in Human Serum and
CSF
BLC
Low QC - Serum (5%) spiked with recombinant analytes, except 6Ckine
CCL19, CCL21 and CXCL13 analytes showing correlation of R2 value >0.5 to ELISA results, with CCL19 exhibiting better sensitivity No correlation between serum LIGHT concentrations derived from ELISA and 10-plex
LOW QC
Verification of BAFF and ITAC/CXCL11 in the 10-plex with known blinded serum samples(n=24) BAFF 8000
ITAC/CXCL11 1200
y = 1.091x R² = 0.830
y = 1.008x R² = 0.956
Biotinylated antibodies binds to cytokines Streptavidin-PE binds Biotinylated antibody to emit Fluorescence Fluorescence measured using Luminex 100 analyser
• • • •
Assay Development
R&D Systems ELISA (pg/mL)
R&D Systems ELISA (pg/mL)
6000 5000 4000 3000 2000
800
2000
3000
4000
5000
RBM 10-plex (pg/mL)
6000
7000
Development Determination of assay performance parameters Evaluation of assay sensitivity Evaluation of assay specificity Optimization of data reduction to express the concentration of analyte in each matrix
Inter CV
Range (pg/mL)
(pg/mL)
Intra CV
Inter CV
8.5-14
11
9%
20%
33-58
42
6%
21%
132-212
169
5%
17%
BAFF
99-121
108
3%
7%
311-417
354
4%
10%
2280-2630
2428
2%
5%
BLC
47-62
54
5%
8%
128-157
142
5%
7%
313-357
333
4%
4%
IP-10
180-230
196
2%
9%
437-486
457
2%
4%
1220-1630
1442
3%
9%
ITAC
32-49
38
7%
15%
121-146
133
2%
7%
1040-1170
1095
3%
4%
MCP-1
32-64
47
16%
20%
106-144
122
6%
9%
431-531
475
4%
7%
MIG
830-1180
969
8%
13%
2670-3560
3119
5%
10%
17300-22500
19760
2%
8%
MIP-3β
462-675
540
7%
11%
761-887
825
2%
5%
1370-1850
1577
6%
9%
SDF-1
77-98
88
7%
8%
295-336
315
3%
4%
1330-1450
1398
3%
3%
Obs
Low QC - Pooled human CSF (5%) spiked with recombinant analytes except IP-10 and MCP-1
CSF
Serum
CSF
Serum
0
200
400
600
800
1000
1200
RBM 10-plex (pg/mL)
CXCL13 was not detected in the ELISA but was detected in the multiplex assay
Nine of ten analytes were scaled up for method validation. Validation was performed for plasma, serum and CSF matrices.
Serum
CSF
Selectivity
Serum
CSF
Serum (n=3)
Plasma (n=3)
CSF (n=3)
Dilution 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16 1:10 1:20 1:40 1:4 1:8 1:16
Exp (pg/mL) 216 104 52 58 29 15 1034 517 258 54 27 13 95 48 58 1079 539 270 1079 539 270 862 431 216 4.6 8325 4163 2081 159 80 40 18350 9175 4588 418 209 104 788 394 197 259 129 65 3848 1924 962 453 226 113
Obs (pg/mL) 203 101 51 58 34 17 1030 549 249 48 22 12 112 63 65 1265 690 350 1265 690 350 862 467 262 3.9 9075 4470 2315 191 89 43 19850 10170 4615 360 200 111 808 357 161 255 133 53 3845 1945 977 492 287 158
Recov 107% 103% 103% 100% 116% 114% 100% 106% 96% 90% 82% 91% 117% 132% N/A 112% N/A N/A 117% 128% 130% 117% 128% 130% 100% 108% 122% 86% N/A N/A 109% 107% 111% 120% 111% 107% 108% 111% 101% 86% 96% 106% 103% 91% 82% 99% 102% 81% 100% 101% 102% 109% 127% 140%
Results: - Clinical samples will be run at an MRD of 1:5 for serum/plasma and 1:2 for CSF, unless a High reading is obtained for an analyte and it requires further dilution (up to the maximal dilution). - Serum and plasma samples can be maximally diluted up to 1:20-1:40 depending upon the analyte, whereas CSF samples can be diluted up to 1:4-1:16.
Spike Level Analytes
1
2
3
1
2
3
1
2
3
6Ckine
85
92
68
118
69
36
96
87
74
BAFF
115
119
102
103
108
97
102
96
104
BLC
122
124
118
94
113
108
81
88
96
IP-10
82
85
83
52
62
59
101
105
104
ITAC
94
92
94
58
66
65
103
104
101
MCP-1
88
89
87
82
97
88
105
104
93
MIG
86
87
77
86
84
86
97
93
100
MIP-3β
132
128
115
92
102
93
81
88
85
SDF-1
63
67
70
41
50
53
116
112
106
Results: CSF recovered all the spiked in concentrations for all the analytes. Serum passed for all recoveries except for SDF-1. Plasma had low recovery when spiking for 6Ckine, IP-10, ITAC, and SDF-1. h, 2- Mid, 3Greater sensitivity in 10-plex compared to ELISA
Serum
Controls were measured in duplicate
Reagent Selection ELISA Kits vs Singleplex and multiplex Luminex reagents Optimization of assay matrix Selection of assay format Selection of assay controls
(pg/mL)
Intra CV
Obs
Medium QC - 1:3.4 dilution of high QC
• Excellent correlation of 10-plex to the ELISA with similar sensitivity Correlation between ELISA and 10-plex derived CCL19 & CXCL13 concentrations in human CSF
Assay range Precision Selectivity Sensitivity LDD & LLOQ Dilution linearity Martix Interference Sample Stability
Range (pg/mL)
High QC - Pooled human CSF (25%) spiked with recombinant analytes except IP-10 and MCP-1
Validation • • • • • • • •
Inter CV
400
0 1000
(pg/mL)
Intra CV
CSF
200
0
6Ckine
Obs
HIGH QC
600
1000 0
Range (pg/mL)
MED QC
CSF
ITAC
•
Precision of CSF-based Controls
MCP-1
•
IP-10
Serum
Add 5ul capture antibody beads + 5ul blocking buffer + 10uL standard/control/sample in 96 well plate Incubate 1 hr at ambient temp Add 10uL biotinylated detection antibody Incubate 1 hr at ambient temp Add 10uL Streptavidin-PE Incubate 1 hr at ambient temp Transfer contents to Pre wetted filter membrane 96 well plate 2x wash with 100uL wash buffer FL intensity read in Luminex 100 analyser Data analysis by Plateviewer software using Best fit option MFI interpolated using standard curve
• • • • • •
CSF
Controls were measured in duplicate
Antibodies for each analyte were purchased and conjugated to beads or biotinylated (detection) using standard protocols. Each bead and detection was screened by Luminex assay with a recombinant protein and serum samples. Pairs were selected based on sensitivity, linearity of dilution and the ability to detect the samples at expected levels, followed by multiplexing in a 10plex assay. Assay verification for MIP-3 beta/CCL19, 6Ckine/CCL21, SDF-1/CXCL12, LIGHT BLC/CXCL13 was performed by running blinded samples in the 9-plex with known values. BAFF and ITAC/CXCL11 were correlated to ELISA. MIG/CXCL9, IP-10/CXCL10 and MCP-1/CCL2 were previously used in multiplex assay and did not undergo further verification. Nine of ten analytes where then scaled up for final validation. The assay was validated for Least Detectable Dose(LDD), lower limit of quantitation (LLOQ), spike-recovery, linearity and precision. Validation was performed for serum and CSF.
Cytokines bind to antibodies
CSF
Serum
Medium QC - Serum (98%) spiked with recombinant analytes except 6Ckine, BAFF, and SDF-1
•
Serum
Serum
High QC - Serum (97%) spiked with all recombinant analytes
Micrspheres coated with antibody
• • •
6Ckine
LOW QC
1000
•
LLOQ (pg/mL)
y = 0.0276x R² = 0.6422
7000
• • •
LDD (pg/mL)
Precision of Serum-based Controls
Assay Format
•
LLOQ (pg/mL)
MIG
•
LDD* (pg/mL)
MIP-3β
•
HQC
SDF-1
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
SD
CSF
BAFF
10-plex Assay Development consisting Cerebrospinal fluid:
n
Serum/Plasma
Mean + 3SD (MFI)
*Least Detectable Dose (LDD) The LDD characterizes the sensitivity of the kit and is determined by adding three standard deviations to the average signal of 20 replicate determinations of standard curve blanks and calculating the corresponding concentration.
Methods •
Mean MFI
Representative Standard Curves
Conclusion Analytical validation supported the use of multiplex assay as a valuable tool for clinical studies to evaluate the pharmacological activity of biotherapeutic in human serum and CSF or monitor disease activity and efficacy trends upon treatment with biotherapeutic. Results from the multiplex assay development matched reference ELISA assay results with equal or greater sensitivity for 9 out of 10 assays The 9-plex assay was shown to be accurate and specific in both serum and CSF matrices The assay is rapid and reproducible and allows for the quantitative measure of 9 analytes in a single 50 µl sample
